Risk assessmentindependent cross-checkfull text checked★ key study I-ABX-01 in-depth review

Antibiotic resistance deaths 1990-2050

Global burden of bacterial antimicrobial resistance 1990-2021: a systematic analysis with forecasts to 2050.

Overview

GBD 2021 Antimicrobial Resistance Collaborators. Lancet 2024 · PMID 39299261 · DOI 10.1016/S0140-6736(24)01867-1

About 1.14 million deaths were caused by resistance in 2021; better care and access to antibiotics could prevent tens of millions of deaths.

Does the study support the specific claim? Partly · Methodological quality: high

Original publication

Plain-language summary

Study explained simply

What did the researchers want to find out?: Track AMR burden.
How was the study done?: GBD modelling.
How many people, animals or samples were studied?: 520 million records.
What was actually found?: 2021: 1.14 million deaths caused by resistance and 4.71 million deaths with a resistant infection; deaths in children under 5 fell by more than half since 1990, but rose by over 80% in people 70 and older.
How large was the observed effect?: 1.91 M forecast for 2050.
What do the results mean in everyday life?: Resistance is rising; access to the right antibiotics matters.
What are the study's limitations?: Model estimates; little data in many poorer countries; forecasts depend on assumptions.
What does it say about Elawady's specific claim?: Supports burden; contradicts refusing antibiotics.
Do the results apply to Germany and Europe?: High.
What can explicitly not be concluded from this study?: Exact counts.

Scientific in-depth review

1. Full reference

Authors: GBD 2021 Antimicrobial Resistance Collaborators
Journal: Lancet
Year: 2024
Identifier: PMID 39299261 DOI 10.1016/S0140-6736(24)01867-1
Verification: verified – PubMed efetch 2026-10-25
Retracted: no
Corrections: none
Extent checked: full text
Conflicts of interest and funding: Funding: Wellcome Trust (A126042) and UK Department of Health and Social Care via the Fleming Fund (R52354 CN001); funders had no role in design, analysis or decision to publish (affiliated coauthors could comment on drafts); forecasts partly supported by the World Organisation for Animal Health (EcoAMR). Numerous individual author declarations (honoraria, advisory roles, industry grants), mostly outside the submitted work.

2. Research question

Trends and forecasts of AMR burden

3. Study design and population

Study design: Risk assessment – Global Burden of Disease (GBD 2021) systematic modelling analysis of bacterial AMR burden 1990-2021 with probabilistic forecasts to 2050 (reference, Gram-negative drug and better-care scenarios)
Population: Global
Country: 204 countries
Period: 1990-2021; forecasts to 2050
Sample size: 520 million individual records or isolates; 19,513 study-location-years; 22 pathogens, 84 pathogen-drug combinations, 11 infectious syndromes; 7 of 204 countries had no data for any component
Inclusion and exclusion criteria: All available multiple-cause-of-death, hospital discharge, microbiology, literature, pharmaceutical sales, antibiotic use, surveillance, linkage and insurance-claims data; estimates for locations without data via covariates and regional patterns

4. Methods and endpoints

Exposure / intervention: Drug-resistant bacterial infections
Comparator: Two counterfactuals: drug-susceptible infection (attributable burden) and no infection (associated burden)
Endpoints: Deaths and DALYs attributable to and associated with AMR, by age, location, pathogen-drug combination
Statistical methods: Five-component GBD modelling (sepsis deaths, syndrome and pathogen fractions, resistance prevalence, relative risk of death) with 95% uncertainty intervals; out-of-sample validity checks; forecasting from aggregate AMR mortality

5. Actual results

Results: 2021: 4.71 million (95% UI 4.23-5.19) associated and 1.14 million (1.00-1.28) attributable deaths. 2019 (re-estimated in this analysis): 4.94 million (4.43-5.45) associated and 1.20 million (1.05-1.35) attributable deaths. 1990: 4.78 million associated, 1.06 million attributable. Attributable death rate fell from 19.8 per 100,000 (1990) to 14.5 (2021). 1990-2021: under-5 attributable AMR deaths fell 60.4% (47.6-73.1); deaths in people >=70 years rose by >80%. MRSA attributable deaths 57,200 (1990) to 130,000 (2021); carbapenem-resistance attributable deaths 127,000 to 216,000. Forecast 2050: 1.91 million (1.56-2.26) attributable and 8.22 million (6.85-9.65) associated deaths; 65.9% of attributable deaths in people >=70. Better-care scenario: 92.0 million (82.8-102.0) deaths averted cumulatively 2025-2050 through better care of severe infections and improved access to antibiotics; Gram-negative drug scenario: 11.1 million (9.08-13.2) averted.
Absolute risks: Population rates: 14.5 (12.7-16.2) attributable AMR deaths per 100,000 in 2021; 8.47% of non-COVID infectious deaths attributable to AMR in 2021
Authors' conclusion: AMR burden is rising among older adults while falling in children under 5; interventions should combine infection prevention, vaccination, minimisation of inappropriate antibiotic use in farming and humans, research into new antibiotics, and better care with improved access to antibiotics.

6. Strengths

Most comprehensive dataset to date with consistent GBD methods
Two counterfactual definitions and explicit uncertainty intervals
Historical trends plus scenario forecasts
Extensive limitations section

7. Weaknesses

Sparse data in many LMICs; 7 countries with no data; heavy reliance on covariates and modelling
Pre-2000 resistance largely modelled
Relative risks of death vulnerable to residual confounding and collider bias (authors' statement)
Selection bias of passive surveillance and hospital microbiology data
COVID-19 disruption of 2020-2021 data
Forecasts assume no new superbugs; scenario assumptions (e.g. 50% reduction in Gram-negative deaths)
Very large author group with heterogeneous declarations

8. Risk of bias

Orienting tool: No standard tool fits (modelling/burden-of-disease study); structured validity judgement
Type of assessment: structured assessment based on the full text by a single reviewer (no formal double review)
Full formal assessment: no
Overall risk of bias: some concerns
Rationale: Single-reviewer structured judgement based on the full text; not formal (needs two independent reviewers).

8. Risk of bias
Domain	Judgement	Rationale
Input data coverage and selection	some concerns	Large but uneven data; LMIC sparsity; hospital and tertiary-care overrepresentation.
Model assumptions	some concerns	Counterfactual definitions, beta-lactam hierarchy, borrowing of relative risks across pathogens.
Confounding in relative-risk estimates	some concerns	Authors acknowledge residual and collider bias in excess-mortality estimates.
Uncertainty quantification	low	95% UIs propagated; out-of-sample validation.
Forecast validity	some concerns	Scenario-dependent; built on historical estimates; cannot anticipate new resistant pathogens.

Full-text review (25 October 2026): Added re-estimated 2019 figures (1.20/4.94 million), 1990 figures and rates, pathogen trends, scenario details and assumptions, funding and role of funders, full author-stated limitations; structured validity domains (overall unchanged: some concerns).

9. Assessment of the statistical interpretation

UIs reflect model uncertainty but not all structural assumptions. Attributable (vs susceptible infection) and associated (vs no infection) deaths answer different questions; 2019 values in this analysis (1.20 and 4.94 million) differ slightly from the earlier GBD 2019 publication.

10. Relevance to the specific claim

C-ABX-02 (update), C-ABX-12 (access to antibiotics saves lives).

11. Independent contrary findings

12. Transferability to Germany and Europe

High.

13. Overall methodological judgement

Authoritative modelled update: AMR is a major and growing cause of death, especially in older adults. The same analysis estimates that better care and improved access to antibiotics could avert 92 million deaths by 2050, which directly contradicts advice to avoid antibiotics altogether.

Multidimensional assessment
Multidimensional assessment
Authenticity of the source	verified
Methodological quality	high
Risk of bias	some concerns
Precision of results	moderate
Directness	direct
Reproducibility	replicated
Consistency with the overall evidence	consistent
Regional transferability	high
Adequacy of the interpretation (by the archive)	not applicable

14. Open questions

Country-level estimates for Egypt; accuracy in data-poor settings.

15. Linked fact checks

Bacterial antimicrobial resistance (AMR) was associated with 4.95 million deaths in 2019, of which 1.27 million were directly attributable to resistance. – Studies supporting the claim
Antibiotics should not be used as a treatment at all; nobody should take an antibiotic for any reason (not even after a tooth extraction); the body can cope on its own. – Studies contradicting or limiting the claim